独立式脑-计算机接口信号处理和识别方法研究

目的为独立式脑-计算机接口(brain-computerinterface,BCI)选择合适的数据预处理、特征提取和模式识别算法,提高系统的识别率.方法分别使用不同的空间滤波算法、特征提取方法和模式识别算法,对同一组BCI数据进行处理,并对结果加以比较分析.结果使用空间滤波器small-laplacian提高信噪比,AR谱估计提取C3和C4导联特定波段能量作为特征,Bayesian判决进行分类,可以取得较好的识别率.结论高信噪比以及合适的特征是提高识别率的关键因素.

Study on Data Processing and Recognition Methods of Independent Brain-Computer Interface

Objective To select appropriate data pre-processing, feature-extraction and pattern recognition algorithms to improve the classification performance of independent Brain-Computer Interface (BCI).Method Using spatial filter, feature extration, and patten recognition algorithms, we processed the same BCI data set, compared and analyzed the performance of the results.Result Satisfying result can be achieved with small-laplacian spatial filter to improve SN rate, AR spectrum estimation to extract specified frequency band of channel C3 and C4 as features, and Bayesian decision rule to classify.Conclusion High SN rate and suitable features are the keys to improve classification accuracy.